KR20070075227A - Flag encoding method, flag decoding method, and apparatus thereof - Google Patents

Abstract

A flag encoding method and apparatus, and a flag decoding method and apparatus are provided to efficiently encode various flags used in a video codec in consideration of spatial correlation. A flag encoding method includes a step of collecting flag values allocated to respective blocks according to spatial correlation of the blocks to generate a flag bit stream(20), a step of splitting the bit stream into groups each having a predetermined size, a step of setting a pattern bit which represents whether bits included in the split bit stream are all zero, and a step of entropy-encoding the split bit stream according to the set pattern bit.

Description

Flag encoding method, flag decoding method, and apparatus using the method

1 is a diagram illustrating macroblocks constituting one frame.

2 illustrates spatial associations between specific flags included in each macroblock header.

3 illustrates a concept of entropy coding of values of the same flags.

4 is a block diagram showing a configuration of a flag encoding apparatus according to an embodiment of the present invention.

5 is a block diagram showing the configuration of a flag decoding apparatus according to an embodiment of the present invention.

6 is a flowchart illustrating a flag encoding method according to an embodiment of the present invention.

7 illustrates the concept of various prediction techniques used in SVCs.

8 is a block diagram showing a configuration of a flag encoding apparatus according to another embodiment of the present invention.

9 is a flowchart illustrating a flag encoding method according to another embodiment of the present invention.

10 is a flowchart illustrating a flag decoding method according to an embodiment of the present invention.

11 is a block diagram showing a configuration of a flag decoding apparatus according to another embodiment of the present invention.

12 is a flowchart illustrating a flag decoding method according to another embodiment of the present invention.

FIG. 13 is a block diagram illustrating a configuration of a video encoder to which the flag encoding apparatus of FIG. 4 or 8 may be applied.

14 is a block diagram illustrating a configuration of a video decoder to which the flag decoding apparatus of FIG. 5 or 11 may be applied.

(Symbol description of main part of drawing)

100 and 190: flag encoding apparatus 101: flag reading unit

105, 260: calculator 110: flag combination

120, 210: group size setting unit 130: bit string division unit

140: pattern bit setting unit 150: flag coding unit

160: recording unit 170: buffer

200, 290: flag decoding device 220: pattern bit reader

230: flag decoding unit 240: output buffer

250: flag setter 300: lower layer encoder

305: downsampling unit 310, 410: prediction unit

320, 420: conversion unit 330, 430: quantization unit

340 and 440: entropy encoding unit 350 and 450: flag setting unit

360: Mux 400: Current Layer Encoder

500: Video Encoder 600: Lower Layer Decoder

610, 710: entropy decoding unit 620, 720: inverse quantization unit

630, 730: inverse transform unit 640, 740: inverse predictor

650: Demux 700: Current Layer Decoder

800: video decoder

The present invention relates to a video compression technique, and more particularly, to a flag coding method and apparatus for efficiently coding using spatial relationships between various flags included in a macroblock.

As information and communication technology including the Internet is developed, not only text and voice but also video communication are increasing. Existing text-oriented communication methods are not enough to satisfy various needs of consumers. Accordingly, multimedia services that can accommodate various types of characteristics such as text, video, and music are increasing. Multimedia data has a huge amount and requires a large storage medium and a wide bandwidth in transmission. Therefore, in order to transmit multimedia data including text, video, and audio, it is essential to use a compression coding technique.

The basic principle of compressing data is to eliminate redundancy in the data. Spatial overlap, such as the same color or object repeating in an image, temporal overlap, such as when there is almost no change in adjacent frames in a movie frame, or the same note over and over in audio, or high frequency of human vision and perception Data can be compressed by removing the psychological duplication taking into account the insensitive to. In a general video coding method, temporal redundancy is eliminated by temporal filtering based on motion compensation, and spatial redundancy is removed by spatial transform.

The result of removing the duplication of data is again loss-coded according to a predetermined quantization step through a quantization process. The quantized result is finally losslessly coded through entropy coding.

Entropy coding techniques currently used in the H.264 standard include Context-Adaptive Variable Length Coding (CAVLC), Context-Adaptive Binary Arithmetic Coding (CABAC), and Exp_Golomb (Exponential Golomb). Table 1 below shows an entropy encoding scheme used for each parameter to be coded in the H.264 standard.

According to Table 1, when the entropy_coding_mode flag is 0, a macroblock type indicating whether the macroblock is an inter prediction mode or an intra prediction mode, and a macroblock pattern indicating the type of a subblock constituting the macroblock ( A macroblock pattern, a quantization parameter which is an index for determining a quantization step, a reference frame index indicating the number of frames referred to in the inter prediction mode, and a motion vector are encoded by Exp_Golomb. Residual data representing the difference between the original image and the predicted image is encoded by CAVLC.

On the other hand, when the entropy_coding_mode flag is 1, all the parameters are coded by CABAC.

CABAC shows good performance on parameters with high complexity. Therefore, Variable Length Coding (VLC) based entropy coding such as CAVLC is set as the basic profile.

Meanwhile, in H.264 or multi-layer-based scalable video coding (called a scalable edition (H.264SE)), there are various flags related to whether inter-layer information is used. These flags are set for each slice, for each macroblock, or for each sub-block. Therefore, the overhead increased by the flag in video coding cannot be ignored.

However, at present, unlike the texture data and the motion data, the flags are encoded or not encoded at all without considering spatial correlation.

SUMMARY OF THE INVENTION The present invention has been devised in view of the above-described needs, and a technical object of the present invention is to provide a method and apparatus for efficiently encoding various flags used in a video codec in consideration of spatial correlation.

The object of the present invention is not limited to the above-mentioned object, and other objects not mentioned will be clearly understood by those skilled in the art from the following description.

In order to achieve the above technical problem, a method of encoding a flag used in the coding of a video frame consisting of a plurality of blocks, (a) collecting the flag value assigned to each block according to the spatial relevance of a predetermined block flag Generating a bit string; (b) dividing the bit string into a predetermined group size; (c) setting pattern bits indicating whether all bits included in the divided bit streams are zero; And (d) entropy encoding the divided bit strings according to the set pattern bits.

In order to achieve the above technical problem, a method of encoding a flag used in coding a video frame composed of a plurality of layers, comprising: (a) a first flag value for a block of a current layer and a lower level corresponding to the block; Performing an exclusive OR operation on the second flag value for the block of the layer; (b) generating a flag bit string by collecting the operation result values according to the spatial relevance of the blocks of the current layer; (c) dividing the bit string into a predetermined group size; (d) setting pattern bits indicating whether all bits included in the divided bit streams are zero; And (e) entropy encoding the divided bit strings according to the set pattern bits.

In order to achieve the above technical problem, a method of decoding a flag used in coding a video frame composed of a plurality of blocks, comprising: (a) a pattern indicating whether all bits included in a predetermined bit string from a video stream are 0; Reading the bits; (b) setting the flag bit string corresponding to the group size to 0 when the pattern bit has the first bit; (c) if the pattern bit has a second bit, reading an encoded flag bit string for as many blocks as the group size from the video stream; And (d) entropy decoding the read flag bit string.

In order to achieve the above technical problem, an apparatus for encoding a flag used in the coding of a video frame consisting of a plurality of blocks, the flag bit string is collected by collecting the flag value allocated to each block according to the spatial relevance of a predetermined block Means for generating; Means for dividing the bit string into predetermined group sizes; Means for setting pattern bits indicating whether all bits included in the divided bit strings are zero; And means for entropy encoding the divided bit strings according to the set pattern bits.

In order to achieve the above technical problem, an apparatus for encoding a flag used in the coding of a video frame consisting of a plurality of layers, the first flag value for the block of the current layer and the lower layer block corresponding to the block; Means for performing an exclusive OR operation on the second flag value for; Means for gathering the operation result values according to the spatial relevance of the blocks of the current layer to generate a flag bit string; Means for dividing the bit string into predetermined group sizes; Means for setting pattern bits indicating whether all bits included in the divided bit strings are zero; And means for entropy encoding the divided bit strings according to the set pattern bits.

In order to achieve the above technical problem, an apparatus for decoding a flag used in coding a video frame composed of a plurality of blocks, the pattern bit indicating whether all bits included in a predetermined bit string are 0 from a video stream. Means for doing so; Means for setting a flag bit string corresponding to a group size to 0 when the pattern bit has a first bit value; Means for reading an encoded flag bit string for blocks as many as the group size from the video stream when the pattern bit has a second bit value; And means for entropy decoding the read flag bit string.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. Advantages and features of the present invention and methods for achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but will be implemented in various forms, and only the present embodiments are intended to complete the disclosure of the present invention, and the general knowledge in the art to which the present invention pertains. It is provided to fully convey the scope of the invention to those skilled in the art, and the present invention is defined only by the scope of the claims. Like reference numerals refer to like elements throughout.

Scalable video coding has significantly increased the amount of syntax compared to conventional MPEG-4 or H.264, where each syntax itself is one bit or a few bits long, but neglects the overhead of syntax if it is based on the entire video stream. Difficult to do The present invention focuses on reducing the overhead due to the flags by collecting the same flags (syntaxes) existing for each block (slice, macroblock, or subblock) in video coding at once.

1 to 3 are diagrams for explaining the basic concept of the present invention.

In video coding, the process is generally performed in units of macroblocks of 16 × 16 pixels. As shown in FIG. 1, one frame or slice is divided into a plurality of macro blocks (MB _n ; n is a natural number). Each macroblock is a video stream through a predetermined lossy encoding and lossless encoding process. For example, the encoding order of the macroblocks may be in a zigzag scan order as shown in FIG.

The macroblocks MB _n encoded as described above are collected after a header (macroblock header) is added to the head to form one video stream 20 as shown in FIG. 2. The headers contain various flags (A_flag _n , B_flag _n, etc.). However, although each header includes the same flag, the value of the flag may be different. However, macroblocks of spatially adjacent positions have spatial correlations or similarities with each other. Thus, the same flag between adjacent macroblocks is likely to have the same value. For example, a coded block pattern (cbp) flag indicating whether a corresponding macroblock is encoded has a large spatial correlation. Such a cbp flag is defined in the H.264 standard and the SVC standard by a parameter called g_aucCbpInter for an inter macroblock and a parameter called g_aucCbpIntra for an intra macroblock.

With this in mind, the present invention proposes encoding after collecting values (bit strings) of the same flags. For example, as shown in FIG. 3, the values A_flag ₁ , A_flag ₂ ,..., A_flag _n are collected and encoded, and the values B_flag ₁ , B_flag ₂ , ..., B_flag _n ), and then entropy coding (lossless coding). As such, when the values of the same flag are collected and encoded, it may be easily predicted that the compression efficiency is higher than that of the flag values separately encoded due to the similarity of the values.

Hereinafter, a description will be given of which algorithm entropy-encodes the collected flag values (hereinafter referred to as "flag bit string") in detail. 4 is a block diagram showing a configuration of a flag encoding apparatus 100 according to an embodiment of the present invention.

The flag encoding apparatus 100 includes a group size setting unit 120, a flag combining unit 110, a bit string dividing unit 130, a pattern bit setting unit 140, a flag coding unit 150, and a recording unit 160. It may include, and may further include a buffer 170.

The flag combination unit 120 collects values of a specific flag for each macroblock and generates a flag bit string as shown in FIG. 3. The flag generally means a flag represented by one bit, but may be a flag expressed by two or more bits.

The group size setting unit 110 sets a group size (gs) which is a unit for dividing a flag bit string. The group size is represented by the number of bits, and may be, for example, 4, 8, or 16. The larger the number of zeros in the distribution of the specific flag value is, the larger the group size is set. The group size may be set in units of any number of macroblocks, and may be set for each slice or frame composed of a plurality of macroblocks. In this case, the gs value may be recorded in the slice header or the frame header.

The bit string dividing unit 130 divides the flag bit string input from the flag combining unit 110 according to the group size (gs value) set by the group size setting unit 110. The bit string divided by the group size unit is input to the pattern bit setting unit 140.

The pattern bit setting unit 140 sets a predetermined bit (hereinafter, referred to as a pattern bit pb) to 1 so as to undergo entropy encoding when all of the divided bit strings are 0. Otherwise, entropy encoding is performed. The pattern bit is set to 0 so as not to go through.

When the pattern bit is set to 0, that is, when there is a non-zero value among the divided bit strings, the flag coding unit 150 performs entropy encoding (lossless encoding) on the divided bit strings. As the entropy coding, techniques known in the art, such as variable length coding and arithmetic coding, may be used. In addition, although it is also possible to entropy-encode the divided bit strings, entropy encoding may be performed by collecting all the divided bit strings including non-zero values again.

The meaning when the pattern bit is 0 and the meaning when 1 may be selected opposite to each other. Therefore, in the present specification, one of two values that can be represented by one bit may be represented by a first bit and the other by a second bit.

The recording unit 160 records the group size gs and the pattern bit pb in the video stream. The group size gs may be recorded in units of slices or frames, and the pattern bit pb may be recorded for each macroblock equal to the group size gs. For example, when the group size is 4, the pattern bit may be recorded in the header of the first macroblock among the four macroblocks. In addition, when pb is 0, the recording unit 160 also records entropy coding results provided from the flag coding unit 150. The entropy encoding result may be recorded in the header of the first macroblock like pb. If entropy encoding is performed at the same time by collecting all of the divided bit streams in the flag coding unit 150, the entropy encoding result may be recorded in a slice header or a frame header.

In the above description, an example of recording and transmitting the set group size gs has been described. However, if the group size gs can be calculated using information common to the encoding apparatus and the decoding apparatus, the group size gs may be calculated. You will not need to send it.

As another embodiment of the flag encoding apparatus 100, the group size setting unit 120 may check the distribution of the flag bit strings related to the previously generated macroblocks so that the group size may be variably changed. For example, the group size setting unit 120 may select a recent predetermined number of bits among the accumulated flag bit strings, and set the group size according to a relationship between a ratio occupied by zero and a threshold value.

As a more detailed example, the group size setting unit 120 sets the group size to 1 when the ratio of 0 among the selected bits is less than 1/3, and sets the group size when the ratio of 0 is greater than 4/5. Set the size to 8. Otherwise, the group size is set to four.

Since the accumulated bits are information that can be commonly known not only in the flag encoding apparatus but also in the corresponding flag decoding apparatus, the recording unit 160 does not need to record the set group size in the video stream.

5 is a block diagram showing the configuration of a flag decoding apparatus 200 according to an embodiment of the present invention.

The flag decoding apparatus 200 may include a group size setting unit 210, a pattern bit reading unit 220, a flag decoding unit 230, an output buffer 240, and a flag setting unit 250. .

The group size setting unit 210 sets a group size by reading group size (gs) information included in a video stream, or sets a group size through distribution of a reconstructed flag value provided from an output buffer 240. . The former is a case where the group size is recorded and transmitted by the flag encoding apparatus 100, and the latter is a case where the group size is set and not transmitted by the flag encoding apparatus 100 according to the bit distribution. In the latter case, the group size setting unit 210 calculates the group size in the same manner as the group size setting unit 120 of the flag encoding apparatus 100.

The pattern bit reading unit 220 reads the value of the pattern bit pb included in the video stream, and if the value is 1, notifies the flag setting unit 250, and if the value is 0, notifies the flag decoding unit 230 ( notify).

When the pb is notified by the pattern bit reading unit 220 as 0, the flag decoding unit 230 encodes a coded flag bit sequence for blocks of the set group size from the video stream (in the flag encoding apparatus 100). Entropy coded result) is read and entropy decoded. The entropy decoding process is performed as the inverse of the entropy coding process performed by the flag coding unit 150.

The flag setting unit 250 sets all of the corresponding flags corresponding to the group size to 0 when the pb is notified as 1 from the pattern bit reading unit 220.

The output buffer 240 temporarily stores the entropy decoded value in the flag decoding unit 230 and a value set in the flag setting unit 250. The stored values are output from the output buffer 240 as a restored flag.

6-9 illustrate flowcharts in accordance with embodiments of the present invention. 6 is a flowchart illustrating a flag encoding method according to an embodiment of the present invention.

First, the group size setting unit 120 sets a group size gs, which is a unit for dividing a flag bit string (S11). The group size can be 4, 8, or 16, for example. The bit string dividing unit 130 divides the flag bit string by the group size unit (S12).

Then, the pattern bit setting unit 140 determines whether the divided bit strings are all zero (S13), and if so, sets the pattern bit pb to 1 (S16), and if not, the pattern bit pb ) Is set to 0 (S14).

The flag coding unit 150 entropy-codes the divided bit string when the pattern bit is set to 0 (S15). In this case, the flag coding unit 150 may entropy-encode the divided bit strings, or may entropy-encode all of the divided bit strings including non-zero values at once.

Finally, if the above steps have been performed for all macroblocks in the slice or frame (YES in S17), the process ends; otherwise, the process proceeds to step S12.

Meanwhile, when a specific flag has only spatial correlation as shown in FIG. 2, a flag encoding algorithm as shown in FIG. 6 may be applied. However, in a multi-layer video coding scheme such as SVC, a specific flag may have hierarchical as well as spatial correlation. For example, the residual prediction flag (residual_prediction_flag), the intra base flag (intra_base_flag), the motion prediction flag (motion_prediction_flag), and the base mode flag (base_mode_flag) may have hierarchical associations, and thus, any macroblock of the current layer may be included. The flag value and the flag value of the macroblock of the lower layer corresponding thereto are likely to have the same value.

The residual prediction flag is a flag indicating whether to use the residual prediction. Residual prediction is a technique for reducing inter-layer redundancy of residual signals by predicting a residual signal of a certain layer using a residual signal of a corresponding lower layer. The lower layer is another layer which is referred to for efficiently encoding a certain layer as an arbitrary layer, and is not limited to the first layer and does not necessarily mean a lower layer.

Whether to use such residual prediction is indicated by the residual prediction flag and passed to the video decoder end. If the flag is 1, the residual prediction is used. If the flag is 0, the flag is not used. In general, a flag of 1 indicates that the feature is enabled, and 0 means that the feature is disabled.

The intra base flag is a flag indicating whether intra base prediction is used. In the current scalable video coding draft, as well as inter prediction (①) and intra prediction (②) used in the existing H.264, as shown in FIG. 7, the image of the lower layer can be encoded by predicting a frame in the current layer. It also supports intra-base prediction (③), which reduces data. In the draft, intra base prediction is treated as a kind of intra prediction. In the intra prediction, if the intra base flag is 0, conventional directional intra prediction is indicated, and if the flag is 1, intra base prediction is indicated.

The motion prediction flag is a flag indicating whether to use a motion vector of the same layer or a motion vector of a lower layer in predicting a motion vector of the current layer to obtain a motion vector difference (MVD). . If the flag is 1, it means that the motion vector of the lower layer is used, and if it is 0, it means that another motion vector of the same layer is used.

The base mode flag is a flag indicating whether to use motion information of a lower layer in displaying motion information of a current layer. If the base mode flag is 1, the motion information of the lower layer is used as the motion information of the current layer as it is, or a value refined somewhat from the motion information of the lower layer is used. If the base mode flag is 0, the motion information of the current layer is separately searched and recorded regardless of the motion information of the lower layer. The motion information includes a macroblock type (mb_type), a reference direction (forward, reverse, bidirectional, etc.) of a picture in inter prediction, and a motion vector.

In any case, the flags having inter-layer correlations have a high probability of having the same value, and thus, when coding the flags for the macroblocks of the current layer, the coding efficiency can be improved. Accordingly, in another embodiment of the present invention, instead of directly coding a flag value for a current layer macroblock, it is proposed to first perform an exclusive OR between the inter-layer flag values and then code the result value.

8 is a diagram illustrating a configuration of a flag encoding apparatus 190 according to another embodiment of the present invention. Comparing FIG. 8 with FIG. 4, the flag reader 101 and the calculator 105 are further included.

The flag reading unit 101 reads a flag F _C for the macroblock of the current layer and a flag F _B for the macroblock of the lower layer corresponding to the macroblock. The reading of the flag can be known, for example, by parsing the macroblock header, or by reading the variable value loaded onto the memory.

The calculating unit 105 performs an exclusive OR operation on the F _C and F _B. As a result, the probability of outputting 0 increases due to hierarchical correlation, and in this case, the efficiency of entropy coding may be increased. Since the result of the calculation in the calculator 105 is input to the flag combination block 110 and the same as in FIG. 4, a description thereof will be omitted.

9 is a flowchart according to the embodiment of FIG. 8.

First, the flag reading unit 101 reads F _C and F _B (S19), and the calculating unit 105 performs an exclusive OR operation by inputting the read F _C and F _B (S20).

Next, the group size setting unit 120 sets a group size gs (S21), and the bit string dividing unit 130 divides the bit string formed by the operation result R _C into the group size unit. (S22).

The pattern bit setting unit 140 determines whether the divided bit strings are all zero (S23), and if so, sets the pattern bit pb to 1 (S26), and if not, sets the pattern bit pb. Set to 0 (S24).

The flag coding unit 150 entropy-codes the divided bit string when the pattern bit is set to 0 (S25). Finally, if the above steps have been performed for all macroblocks in the slice or frame (YES in S27), the process ends; otherwise, the process proceeds to step S22.

10 is a flowchart illustrating a flag decoding method according to an embodiment of the present invention.

First, the group size setting unit 210 sets a group size gs (S31). At this time, the group size setting unit 210 reads the group size (gs) information included in the video stream to set the group size, or the group size through distribution of pre-restored flag values provided from the output buffer 240. Can be set. In the latter case, the group size setting unit 210 calculates the group size in the same manner as the group size setting unit 120 of the flag encoding apparatus 100.

The pattern bit reading unit 220 reads the value of the pattern bit pb included in the video stream (S32), and determines whether the value is 0 or 1 (S33).

If the determination result is 1 (YES in S33), the flag setting unit 250 sets all the flags corresponding to the group size to 0 (S35).

As a result of the determination, if the value is 0 (NO in S33), the flag decoding unit 230 encodes a coded flag bit string (entropy coded by the flag encoding apparatus 100) for blocks corresponding to the set group size from the video stream. Result), and entropy decrypted it (S34). The entropy decoding process is performed as the inverse of the entropy encoding process performed by the flag coding unit 150.

Finally, if the above steps have been performed for all macroblocks in the slice or frame (Yes S36), the process ends; otherwise, the process proceeds to step S32.

On the other hand, when the flag is encoded in consideration of the inter-layer correlation as shown in FIG. 8, some components must be added to FIG. 5 for decoding. 11 is a diagram illustrating a configuration of a flag decoding apparatus 290 according to another embodiment of the present invention.

Comparing FIG. 11 with FIG. 5, the operation unit 260 is further included. Since the operation before the calculator 260 is the same as that of FIG. 5, it is omitted. The value output from the output buffer 240 is R _{C, which} is the result of the calculation of the exclusive OR of F _C and F _B. The operation unit 260 restores F _C , which is a specific flag for the macroblock of the current layer, by performing an exclusive OR on F _B and R _C.

12 is a flowchart according to the embodiment of FIG. 11.

First, the group size setting unit 210 sets a group size gs (S41). At this time, the group size setting unit 210 sets the group size by reading the group size (gs) information included in the video stream, or sets the group size through a distribution of previously restored flag values provided from the output buffer 240. Can be set.

The pattern bit reader 220 reads the value of the pattern bit pb included in the video stream (S42), and determines whether the value is 0 or 1 (S43).

As a result of the determination, if the value is 1 (YES in S43), the flag setting unit 250 sets all of R _C corresponding to the group size to 0 (S46).

As a result of the determination, if the value is 0 (NO in S43), the flag decoding unit 230 reads the encoded data included in the video stream, that is, R _C as much as the group size, and entropy decodes it (S44).

Operation unit 260 performs an exclusive OR with respect to the flag value F of the lower layer corresponding to the _C and F _B, the entropy-decoded in the R _C R _C, or the step S44 is set to 0 in the step S46 (S45).

Finally, if the above steps have been performed for all macroblocks in the slice or frame (Yes 46), the process ends, otherwise proceeds to step S42.

FIG. 13 is a block diagram illustrating a configuration of a multi-layer based video encoder 500 to which the flag encoding apparatuses 100 and 190 of FIG. 4 or 8 may be applied.

The original video sequence is input to the current layer encoder 400, downsampled by the downsampling unit 305 (only when there is a change in resolution between layers), and then input to the lower layer encoder 300.

The predictor 410 obtains a residual signal by differentiating an image predicted by a predetermined method in the current macroblock. The prediction methods include directional intra prediction, inter prediction, intra base prediction, and residual prediction.

The transform unit 420 converts the obtained residual signal using a spatial transform technique such as DCT or wavelet transform to generate transform coefficients.

The quantization unit 430 quantizes the transform coefficients in a predetermined quantization step (the larger the quantization step, the higher the loss or compression ratio of data) and generates quantization coefficients.

The entropy encoder 440 losslessly encodes the quantization coefficients and outputs a current layer video stream.

The flag setting unit 450 sets a flag from information obtained in various steps. For example, the residual prediction flag (residual_prediction_flag), the intra base flag (intra_base_flag), and the like are set through the information obtained by the prediction unit 410. The flags F _C of the current layer set as described above are input to the flag encoding apparatus 190.

As in the current layer encoder 400, the lower layer encoder 300 also has the same function as the predictor 310, the transformer 320, the quantizer 330, the entropy encoder 340, and the flag setter ( 350). The entropy encoder 340 outputs a lower layer video stream to a mux (multiplexer) 360, and the flag setting unit 350 provides the flag layers F _B of the lower layer to the flag encoding apparatus 100.

The mux 360 combines the current layer video stream and the lower layer video stream to generate a video stream VS, and provides the generated video stream to the flag encoding apparatus 100.

Flag encoding apparatus 100 encodes the F _C using the relationship between the supplied F _B, F _C, and the encoded F _C and by inserting the video stream provided above the provided F _B, the final video stream (final BS )

Of course, the flag encoding apparatus 190 may be replaced with the flag encoding apparatus 100. In this case, since the correlation between layers is not used, F _C is entropy encoded using only spatial correlation regardless of F _B.

FIG. 14 is a block diagram illustrating a configuration of a multi-layer based video decoder 800 to which the flag decoding apparatuses 200 and 290 of FIG. 5 or 11 may be applied.

The input final video stream VS is input to the flag decoding apparatus 200 and the demux (demultiplexer) 650. Demux 650 separates the final video stream into a current layer video stream and a lower layer video stream and provides them to the current layer encoder 700 and the lower layer decoder 600, respectively.

The entropy decoder 710 restores quantization coefficients by performing lossless decoding in a manner corresponding to the entropy encoder 440.

The inverse quantization unit 720 inversely quantizes the reconstructed quantization coefficient to the quantization step used in the quantization unit 430.

The inverse transform unit 730 inversely transforms the inverse quantized result by using an inverse spatial transform technique such as inverse DCT transform and inverse wavelet transform.

The inverse predictor 740 obtains the predicted image obtained by the predictor 410 in the same manner, and reconstructs the video sequence by adding the obtained predicted image to the inverse transformed result. In this case, the inverse predictor 740 may use the result decoded by the lower layer decoder 600 in some cases (in the case of intra base prediction, residual prediction, etc.).

As in the current layer decoder 700, the lower layer decoder 600 includes an entropy decoder 610, an inverse quantizer 620, an inverse transform unit 630, and an inverse predictor 640 having the same function. .

Meanwhile, the flag decoding apparatus 290 extracts the encoded values of the flags F _B of the lower layer and the flags F _C of the current layer from the final video stream, and the F _B and the encoded values. Restores the flags F _C of the current layer from.

The extracted flags of the lower layer F _B may be used to perform a corresponding operation in the components 610, 620, 630, and 640 of the lower layer decoder 600, and the restored current layer. The flags F _C may be used to perform the corresponding operation in the components 710, 720, 730, 740 of the current layer decoder 700.

Of course, the flag encoding apparatus 290 may be replaced by the flag encoding apparatus 200. In this case, since the correlation between layers is not used, F _C is entropy decoded using only spatial correlation regardless of F _B.

Until now, each component of FIGS. 4, 5, 8, 11, 13, and 14 is a software (such as a task, a class, a subroutine, a process, an object, a thread of execution, a program) performed in a predetermined area on a memory. software, or a hardware such as a field-programmable gate array (FPGA) or an application-specific integrated circuit (ASIC), or a combination of the above software and hardware. The components may be included in a computer readable storage medium or a part of the components may be distributed and distributed among a plurality of computers.

Although embodiments of the present invention have been described above with reference to the accompanying drawings, those skilled in the art to which the present invention pertains may implement the present invention in other specific forms without changing the technical spirit or essential characteristics thereof. I can understand that. Therefore, it should be understood that the embodiments described above are exemplary in all respects and not restrictive.

According to the present invention, it is possible to improve encoding efficiency of various flags used in a multi-layer scalable video codec.

Claims (23)

A method of encoding a flag used in coding a video frame composed of a plurality of blocks, (a) generating a flag bit string by collecting flag values allocated for each block according to spatial relevance of a predetermined block; (b) dividing the bit string into a predetermined group size; (c) setting pattern bits indicating whether all bits included in the divided bit streams are zero; And (d) entropy encoding the divided bit streams according to the set pattern bits. The method of claim 1, And the block is a slice, a macroblock, or a subblock. The method of claim 1, And the group size is any one of 4, 8, or 16. The method of claim 1, The group size is determined by comparing a threshold with a ratio of 0 included in a predetermined number of bits of a flag bit string previously accumulated. The method of claim 4, wherein The threshold is 1/3 and 4/5. The method of claim 5, If the ratio of zero is smaller than 1/3, the group size is determined as 0. If the ratio of zero is greater than 4/5, the group size is determined as 8. Otherwise, the group size is determined. The flag encoding method is determined as 4. The method of claim 1, wherein the flag is A flag encoding method that is a residual prediction flag or a coded block pattern (cbp) flag. The method of claim 1, The pattern bit is recorded in a header of the block, and the group size is recorded in a slice header. A method of encoding a flag used in coding a video frame composed of a plurality of layers, (a) performing an exclusive OR operation on a first flag value for a block of a current layer and a second flag value for a block of a lower layer corresponding to the block; (b) generating a flag bit string by collecting the operation result values according to the spatial relevance of the blocks of the current layer; (c) dividing the bit string into a predetermined group size; (d) setting pattern bits indicating whether all bits included in the divided bit streams are zero; And (e) entropy encoding the divided bit strings according to the set pattern bits. The method of claim 9, And the block is a slice, a macroblock, or a subblock. The method of claim 9, And the group size is determined by comparing a threshold with a ratio of 0 included in a predetermined number of bits of the flag bit strings. The method of claim 9, wherein the flag is A flag encoding method that is a residual prediction flag. A method of decoding a flag used in coding a video frame composed of a plurality of blocks, (a) reading a pattern bit from the video stream indicating whether the bits included in the predetermined bit string are all zeros; (b) setting the flag bit string corresponding to the group size to 0 when the pattern bit has the first bit; (c) if the pattern bit has a second bit, reading an encoded flag bit string for as many blocks as the group size from the video stream; And (d) entropy decoding the read flag bit string. The method of claim 13, And the block is a slice, a macroblock, or a subblock. The method of claim 13, And said group size is read from said video stream. The method of claim 13, The group size is determined by comparing a threshold with a ratio of 0 included in a predetermined number of bits of a flag bit string entropy decoded in advance. The method of claim 16, The threshold is 1/3 and 4/5. The method of claim 17, If the ratio of zero is smaller than 1/3, the group size is determined as 0. If the ratio of zero is greater than 4/5, the group size is determined as 8. Otherwise, the group size is determined. The flag decoding method is determined as 4. The method of claim 13, wherein the flag is A flag decoding method that is a residual prediction flag or a coded block pattern (cbp) flag. The method of claim 13, and (e) performing an exclusive OR on the flag flag bit string set to 0 in step (b) or the flag bit string entropy decoded in step (d) and the corresponding flag bit string of the lower layer. Decryption method. An apparatus for encoding a flag used in coding a video frame composed of a plurality of blocks, Means for generating a flag bit string by collecting flag values allocated for each block according to spatial relevance of a predetermined block; Means for dividing the bit string into predetermined group sizes; Means for setting pattern bits indicating whether all bits included in the divided bit strings are zero; And And means for entropy encoding the divided bit strings according to the set pattern bits. An apparatus for encoding a flag used in coding a video frame composed of a plurality of layers, Means for performing an exclusive OR operation on a first flag value for a block of a current layer and a second flag value for a block in a lower layer corresponding to the block; Means for gathering the operation result values according to the spatial relevance of the blocks of the current layer to generate a flag bit string; Means for dividing the bit string into predetermined group sizes; Means for setting pattern bits indicating whether all bits included in the divided bit strings are zero; And And means for entropy encoding the divided bit strings according to the set pattern bits. An apparatus for decoding a flag used in coding a video frame composed of a plurality of blocks, Means for reading a pattern bit from the video stream to indicate whether all bits included in the predetermined bit string are zero; Means for setting a flag bit string corresponding to a group size to 0 when the pattern bit has a first bit value; Means for reading an encoded flag bit string for as many blocks as the group size from the video stream when the pattern bit has a second bit value; And And means for entropy decoding the read flag bit string.

Images